Non-Typical Kernel

nekros · Post by **nekros** » Thu Mar 06, 2008 3:44 pm

Hey,
I want to build a kernel that is not like other hobby OSs out there. Most hobby OSs have all the same way of working in general. Do you have any suggestions for things that have not been tested in kernels?

01000101 · Post by **01000101** » Thu Mar 06, 2008 4:01 pm

I disagree, if you looked around at some of the hobby OS's you will realize that there are alot of variations such as the core kernel type: microkernel, monolithic, exokernels... and with very different tasking models.

If you want to create something unique, you should try making one of the 'normal' kernels and then find out what is inefficient about it and change it.

You could even go as far as to create a new kernel structure. or use 64-bit to your advantage, or create new memory managment models... or etc... you get the point?

I have created many experimental kernels trying out new ideas that pop into my head, and some have worked out will, such as implementing direct memory access throughout a kernel to cutdown on overhead.

jerryleecooper · Post by **jerryleecooper** » Thu Mar 06, 2008 8:45 pm

My suggestion is, stay away from unix and unix-like systems. do your own system. create your own system, no unix!

01000101 · Post by **01000101** » Thu Mar 06, 2008 9:19 pm

I agree. Copy-and-paste OSes are nothing (in my opinion) to be proud of.

JamesM · Post by **JamesM** » Fri Mar 07, 2008 2:38 am

I agree. Copy-and-paste OSes are nothing (in my opinion) to be proud of.

Should I even dignify this with an answer? If I did, it would go something along the lines of "go f**k yourself". I mean seriously, that is a 110% flamebait post.

You consider a POSIX compliant kernel a "Copy-and-paste OS'? Get your facts right and remove your head from your own arse before you start posting clueless shite on an otherwise intelligent forum.

pcmattman · Post by **pcmattman** » Fri Mar 07, 2008 2:59 am

jerryleecooper wrote:My suggestion is, stay away from unix and unix-like systems. do your own system. create your own system, no unix!

Have fun porting stuff - who wants to reinvent the wheel anyway?

01000101 · Post by **01000101** » Fri Mar 07, 2008 4:16 am

I fail to see the point where taking someone elses code, re-posting it on the web, and calling it your own is dignified?

Also, get your facts right before accusing me of having my facts wrong. I was mearly stating that re-inventing the wheel is nothing to be proud of... with that said, taking what has already been invented and modifying it in some way can create great things. I never even mentioned the POSIX standard in my previous post btw.

JamesM · Post by **JamesM** » Fri Mar 07, 2008 4:31 am

01000101 wrote:I fail to see the point where taking someone elses code, re-posting it on the web, and calling it your own is dignified?

Also, get your facts right before accusing me of having my facts wrong. I was mearly stating that re-inventing the wheel is nothing to be proud of... with that said, taking what has already been invented and modifying it in some way can create great things. I never even mentioned the POSIX standard in my previous post btw.

Right, let's look at this logically. You agreed with the above post ("I agree."), which was about not creating a UNIX compatible system. The post above was written in such a way as to merely convey a personal opinion and conveys no offence that I can see.

Copy-and-paste OSes are nothing (in my opinion) to be proud of.

I fail to see the point where taking someone elses code, re-posting it on the web, and calling it your own is dignified?

This insinuates one main thing - that you are of the opinion that creating a UNIX compliant operating system consitutes a copy-paste OS. Which I heavily dispute and which offends me personally, as I am creating such an OS.

As to your reply, "I was mearly stating that re-inventing the wheel is nothing to be proud of... " - This again is an opinion that I feel you should keep locked up away from public forums. Many people, myself included, often attempt to "einvent the wheel" in an attempt to further understand the methods behind something. Linux wouldn't have been written if Linus hadn't "reinvented the wheel" that was MINIX.

I never even mentioned the POSIX standard in my previous post btw.

It was implied, as your post follows on from and agrees to the previous post, which relates to UNIX systems and therefore the POSIX standard.

In case you hadn't realised yet, you have actually managed to offend me by your comments.

Solar · Post by **Solar** » Fri Mar 07, 2008 4:35 am

Calm down, guys.

01000101, I read your post (of 03:19 UTC) as that you consider all "unix-like systems" (jerryleecooper) to be "copy-and-paste OSes" (you).

Copy & paste sucks, no matter whether it's copied from Unix or elsewhere.

But no matter what you (or I) think about the merits and deficiencies of unix-like systems, writing one on your own (i.e., by taking the specs and implementing them) would surely be a remarkable feat, and nothing to be sneered at.

JamesM did express this more violently than necessary, but in essence I agree with him.

zaleschiemilgabriel · Post by **zaleschiemilgabriel** » Fri Mar 07, 2008 5:47 am

I want to do the same... Here's my idea of a kernel:
* written completley in assembly
* everything is 64 bit (except initialization)
* multiprocessing
* ACPI support - this is most important, because for some reason I think it can be used for a lot of things, although the ACPI spec is 600 pages...
So if you want a challenge, I think you should try implementing full ACPI support.

Brendan · Post by **Brendan** » Fri Mar 07, 2008 12:55 pm

Hi,

zaleschiemilgabriel wrote:So if you want a challenge, I think you should try implementing full ACPI support.

That is indeed a mighty challenge!

AFAIK there's only 2 "people" who have ever supported ACPI - Microsoft (Windows) and Intel (Open/Free/NetBSD, Linux, Solaris, etc - see the ACPI CA web page for information on Intel's "ACPI Component Architecture"). Something tells me you need to be one of the largest computer related companies in the world (and also on the committee responsible for writing the ACPI specification) to attempt to support ACPI yourself...

@nekros: If you don't want to build something like a hobby OS, try writing a high quality "commercial" OS instead. The first step would be to write a hobby OS to get experience...

Cheers,

Brendan

lukem95 · Post by **lukem95** » Fri Mar 07, 2008 1:10 pm

write a different type of OS then, not a standard desktop OS.

how about an live cd that you boot into, with a simple GUI and will overclock and benchmark test your pc at every interval automatically?

i'd love to see that made. I thought about trying to start that project once, but im busy on my kernel.

zaleschiemilgabriel · Post by **zaleschiemilgabriel** » Fri Mar 07, 2008 5:35 pm

lukem_95 wrote:how about an live cd that you boot into, with a simple GUI and will overclock and benchmark test your pc at every interval automatically?

Live cd: It's been done many times.
What you're describing is not an OS, luke. Sound more like a utility to me.

binutils · Post by **binutils** » Fri Mar 07, 2008 5:51 pm

IMHO, different kernel in different hardware
say(semi-non-sense, just example), movitz in FPGA-mmix -> (os name) in (different hardware).

artu wrote: Those who do not understand Unix are condemned to reinvent it, poorly.
-- Henry Spencer Usenet signature, November 1987

zaleschiemilgabriel · Post by **zaleschiemilgabriel** » Fri Mar 07, 2008 6:06 pm

Just take a look at MenuetOS. It's not Unix and IMO it's not poor.